SPECTRAL DISTRIBUTION OF PROSODIC INFORMATION

Prosodic speech cues for rhythm, stress, and intonation are related pr imarily to variations in intensity, duration, and fundamental frequenc y. Because these cues make use of temporal properties of the speech wa veform they are likely to be represented broadly across the speech spe ctrum. In order to determine the relative importance of different freq uency regions for the recognition of prosodic cues, identification of four prosodic features, syllable number, syllabic stress, sentence int onation, and phrase boundary location, was evaluated under six filter conditions spanning the range from 200-6100 Hz. Each filter condition had equal articulation index (AI) weights, AI = 0.10; p(C)(isolated wo rds) approximate to 0.40. Results obtained with normally hearing subje cts showed that there was an interaction between filter condition and the identification of specific prosodic features. For example, informa tion from high-frequency regions of speech was particularly useful in the identification of syllable number and stress, whereas information from low-frequency regions was helpful in identifying intonation patte rns. In spite of these spectral differences, overall listeners perform ed remarkably well in identifying prosodic patterns, although individu al differences were apparent. For some subjects, equivalent levels of performance across the six filter conditions were achieved. These resu lts are discussed in relation to auditory and auditory-visual speech r ecognition.